THE SPIRITUAL BUT NOT RELIGIOUS: WHO ARE THEY, AND WHO IS MORE LIKELY TO BE ONE? Elizabeth A. Gabhart. Thesis Prepared for the Degree of

Transcription

1 THE SPIRITUAL BUT NOT RELIGIOUS: WHO ARE THEY, AND WHO IS MORE LIKELY TO BE ONE? Elizabeth A. Gabhart Thesis Prepared for the Degree of MASTER OF SCIENCE UNIVERSITY OF NORTH TEXAS May 2015 APPROVED: Gabriel Ignatow, Major Professor Phillip Yang, Committee Member George Yancey, Committee Member Daniel Rodeheaver, Chair of the Department of Sociology Mark Wardell, Dean of the Toulouse Graduate School

2 Gabhart, Elizabeth A. The spiritual but not religious: Who are they, and who is more likely to be one? Master of Science (Sociology), May 2015, 52 pp., references, 64 titles. The spiritual but not religious (SBNR) are a rising social group in America in the past two decades, but social scientists and the general public know quite little about this group. Using the pooled 2006, 2008, 2010, and 2012 GSS data, this study examines who the SBNR are and who is more or less likely to be SBNR controlling for other variables. Descriptive analysis reveals that, compared to the general U.S. adult population, the SBNR group has slightly more males, is slightly younger, has fewer racial minorities, is better educated, and is slightly higher in social class. Additionally, more SBNR are from the Northeast and West than the general population, are slightly more urban, fewer are currently married, fewer have children, more have had homosexual sex, and more were religious Nones when they were 16 years old. Logistic regression analysis of the SBNR finds that, holding other variables constant, Americans who are more educated, live in Northeastern or Western regions, have homosexual sex, or had no religion at age 16 are more likely to be SBNR than their respective counterparts. Those who are racial minorities, live in the South or the Midwest, are currently married, or have children are less likely to be SBNR than their respective counterparts. Gender, age, social class, full-time work status, and metropolitanism of area do not make a significant difference. The implications of the findings for the research of religion and spirituality are discussed.

3 Copyright 2015 By Elizabeth A. Gabhart ii

4 TABLE OF CONTENTS INTRODUCTION... 6 LITERATURE REVIEW... 8 Page Distinguishing Between Spiritual But Not Religious and Religious Nones... 8 Defining Spirituality and Religion Distinct or Overlapping Concepts?... 8 The Spiritual but Not Religious Group Characteristics HYPOTHESES Demographic Variables Socioeconomic Variables Locational Variables Family Life Variables DATA AND METHOD Data Dependent Variable Figure 1. Variations in Spiritual and Religious Orientation Categories Dependent on Operationalization, General Social Survey, U.S. Adults Independent Variables Limitations of Data Methods and Analytical Strategies RESULTS Who Are Spiritual but Not Religious Americans? Table 1. Descriptive Statistics of Spiritual but Not Religious Who Is More or Less Likely To Be Spiritual but Not Religious? Table 2. Predicting Spiritual but Not Religious Orientation Sensitivity Analysis Table 3. Sensitivity Analysis DISCUSSION Demographic Variables Socioeconomic Variables Locational Variables iii

5 Family Life Variables CONCLUSION REFERENCES iv

6 INTRODUCTION The spiritual but not religious (SBNR) are a rapidly growing social category in America. While the majority of Americans claim to be spiritual and religious, a growing minority of them identify themselves as spiritual only (Baker & Smith, 2009a; Dein, 2005; Fuller, 2001; Hout & Fischer 2002; Kosmin, Keysar, Cragun, & Navarro-Rivera, 2009; Marler & Hadaway, 2002; Roof, Greer, Johnson, & Leibson, 1993; Saucier & Skrzypinska, 2006; Streib, 2008). About a decade ago, the SBNR were estimated to account for between 14 percent and 19 percent of the U.S. population (Hout & Fischer, 2002; Marler & Hadaway, 2002; Zinnbauer, 1997). A recent survey by PEW Research Center (PEW Forum, 2012) found that the SBNR group now hovers at around 18 percent of the U.S. population. However, social scientists and the general public know relatively little about this group, and most of the research is dated. Much more research has been devoted to Americans who claim no religious affiliation, known as religious Nones, than to the SBNR. However, a broad examination of all religious Nones combines two distinct groups: those who reject religion but embrace spirituality (SBNR), and those who are neither spiritual nor religious (NSNR). While both of these groups reject religion and therefore share some characteristics, their acceptance or rejection of spirituality divides them into two groups, and should also be considered in discussions of their orientations toward belief, morality, behavior, and organizational belonging. The purpose of this paper is to examine the spiritual but not religious. Specifically, I seek to answer two research questions. First, what are the characteristics of SBNR Americans, and how do their characteristics compare to those of the general population? Second, who is more or less likely to be SBNR, controlling for other characteristics? I use the

7 General Social Survey data to answer these questions. In the remainder of this paper, I define the SBNR, review the literature relevant to the research questions, propose hypotheses for testing, describe the data and methods of analysis, present findings, and discuss their implications. 2

8 LITERATURE REVIEW Distinguishing Between Spiritual But Not Religious and Religious Nones Much recent social science research is devoting to understanding religious Nones, people who, when asked to declare their religion, choose none. Religious None is a broad category of people who reject religion, and this category can be divided into those who embrace spirituality and those who reject spirituality. While not all researchers include spirituality as a specific category of interest when describing religious Nones, most acknowledge that the group is diverse and includes people who have beliefs that can be defined as spiritual. For example, Kosmin et. al. (2009) finds that among religious Nones, a full 24% believe in a higher power and another 27% believe in a personal God. Taken together, these figures mean that 51% of all religious Nones subscribe to ideas which might commonly be called spiritual. This figure which is far lower than the percentage of all adults in the United States who believe in a higher power or a personal God (82%), which leads many scholars to emphasize the lack of belief among Nones. However, in this study, I examine the subcategory of religious Nones who describe themselves as spiritual. Defining Spirituality and Religion Distinct or Overlapping Concepts? Researchers of religion and spirituality disagree about the extent to which spirituality and religion are separate concepts. I begin by describing research which suggests that among Americans, spirituality and religion are distinct concepts, and contrast it with another body of literature which views spirituality and religion as intimately connected, overlapping concepts. 3

9 Finally, I describe the attempts to synthesize these two perspectives and my definition of spiritual but not religious. Spirituality and Religion as Distinct Concepts In academia and in popular culture, religion and spirituality are often assumed to be separate concepts, such that the growth of one necessitates the decline of the other (Ammerman, 2013; Heelas, Woodhead, Seel, Tustin, & Szerszynski, 2005). Many Americans associate religion with orthodoxy and authoritarianism while linking spirituality with a search for meaning, mysticism, and New Age beliefs and practices (Miller & Thoresen, 2003; Pargament, 1999; Pargament, 2007; Zinnbauer et al. 1997). Zinnbauer et al. asked respondents to define the terms spiritual and religious in open-ended questions, and then coded the answers thematically. They found that, in order from most frequent to least frequent themes, spiritual was defined as connectedness with a higher power; personal beliefs; demonstrating God s love to others; attaining a positive inner state; and personal growth. Again, in order from most frequent to least frequent themes, religious was defined as personal belief in a higher power; attendance, membership, and performance of rituals; commitment to dogma; following God s will and demonstrating God s love to others; personal worship practices; and feeling superior to others or avoiding responsibility. Religion is perceived as institutional, while spirituality is perceived as personal and subjective, a finding duplicated among Canadian health care workers (Pesut & Reimer-Kirkham, 2009), African-American women (Mattis, 2000), 4

10 homosexuals surveyed at a gay pride parade (Halkitis, 2009), and Scottish practitioners of alternative spirituality practices (Glendinning & Bruce, 2006). Not only are the concepts of spirituality and religiousness sometimes seen as distinct, but the people who adopt each label are also seen as distinct groups. Traditionally religious people have different dispositions than spiritual but not religious people (Saucier & Skrzypinska, 2006). Traditionally religious people value rituals and rules more, and are more authoritarian, more traditional, less individualistic, and less open to new experiences than people who have what Saucier and Skrzypinska call subjective spirituality. In contrast, people who emphasize spirituality over religiousness, including those who explicitly label themselves as SBNR, are more absorption- and fantasy-prone and value eccentricity and subjective experience more than religious people. Only a few studies exist which specifically identify and examine people who are spiritual but not religious. Fuller s (2001) book provides a qualitative examination of the SBNR. He draws on history to demonstrate that nonreligious spiritual traditions in America are as old as the United States itself, and argues that the success of SBNR philosophy and literature is based in already-existing metaphysical tradition. He contends that eclectic spiritual seekers only belong to religious organizations in as much as they perceive those organizations fulfilling their own individual spiritual searches, rarely out of a sense of obligation to the organization. However, it should be noted that Fuller s argument that spiritual but not religious Americans are largely drawn to New Age and metaphysical beliefs is contested. Supporting Fuller s analysis, SBNR Americans have been shown to be more likely to use alternative medicine such as meditation or spiritual healing than any other spiritual/religious group 5

11 (Ellison, Bradshaw, & Roberts, 2012). However, Pew Forum (2012) has found no quantitative evidence that religious Nones are more drawn toward New Age beliefs than the general public. Perhaps SBNR Americans are more drawn to New Age beliefs and practices than all religiously non-affiliated, a group which includes agnostics and atheists. Alternately, perhaps Fuller s more qualitative research approach unintentionally biased his work toward a particular subset of SBNR Americans who lean toward New Age beliefs more than the average SBNR American. Spirituality and Religion as Overlapping Concepts Many studies treat spirituality and religion as overlapping but not identical concepts (Chatters et al., 2008; Mahoney & Gina, 1999; Saucier & Skrzypinska, 2006). Casey (2013) argues that historically in the Western world, most people were assumed to be religious, while only an elite group of holy individuals ever attained spirituality. For centuries, Europeans considered themselves Christian by virtue of birth, similar to modern citizenship. Social norms in the United States and Europe dictated that everyone engage in religious ritual, especially on important occasions, holidays, and life events. The word spiritual described people who were especially dedicated to religious practice priests, monks, nuns, and saints. To describe someone as spiritual was to acknowledge their higher level of devotion and morality. If this generally-accepted view of religion in the Western world is correct, then historically most Westerners would have called themselves religious, no matter their beliefs, and most would not have identified themselves as spiritual. Historically, the concept of religiousness was broad, and spiritual was a subset of religion. 6

12 In the 1960s, Casey argues, this relationship between religiousness and spirituality began to flip, such that most Westerners might call themselves spiritual, but only a subset of those most likely to regularly engage in religious rituals consider themselves religious. There is no doubt that today, spirituality is generally considered a broader concept than religion, encompassing religion (Hill et al., 2000). People who have little or no interest in deities, religious organizations, or ritual practices may call themselves spiritual to signify their sense of awe in the presence of beauty, a fleeting feeling of connectedness with nature, or a desire to live a purposeful life (Ammerman 2013). This modern definition of spirituality is broader; most Americans feel they can identify with a spiritual existence. According to the view that religion and spirituality are overlapping concepts, the modern definition of religion has shrunk to mean only those people who belong to religious organizations, engage in religious ritual, and frequently engage in faith-based practices. People who are religious are assumed to also be spiritual because the religious rituals and behaviors are thought to be rooted in spirituality (Shahabi et al., 2002). Therefore, modern Americans would define spirituality as a broader concept, with religiousness as a subset of it. Researchers who argue that religion and spirituality are overlapping concepts base this approach largely on evidence that most Americans describe themselves as both religious and spiritual (Chaves, 2011). Americans who describe themselves as religious are more likely than those who are not religious to also describe themselves as spiritual, and vice versa (Shahabi et al., 2002). Given the strong correlation between self-identification as spiritual or religious, 7

13 and the shifting historical overlap of the words, the argument can be made that the modern definitions of spiritual and religious overlap. Synthesis between the Two Views Some recent research aims at understanding how spirituality and religion can be both distinct, and overlapping concepts, focusing on how the word spiritual is contested by different groups. The confusion about how spirituality and religion are defined may be, in part, a result of different definitions used by different groups of Americans. Whether spirituality is viewed as a separate category from religion, or whether it is as seen as simply another dimension of religion depends largely on frequency of religious service attendance and on denominational affiliation (Ammerman, 2013). Americans who attend religious services as often or more than the national average, as well as those who belong to Protestant, Mormon, and evangelical groups, are much more likely to describe spirituality in theistic terms, or in ways that focus on ritual and/or religious institutions. Americans who attend religious services rarely, as well as Jews, neo-pagans, and Catholics, are more likely to use the word spiritual in opposition to religion, as a synonym for individual seeking, mystical experiences, and/or transcendence. The idea of spirituality as an ethical stance, a description of moral behavior rooted in genuine faith, is commonly used by most Americans, across denominational and attendance divides. Although there is broad disagreement among Americans about how to define spiritual and religious, the disagreement is particularly extreme among those who define themselves 8

14 as one or the other. When only spiritual but not religious respondents are asked to define spirituality for themselves, they associate the word with a non-material dimension of existence, personal experience, a universal core of all religions, and a belief in a higher presence (Streib, 2008). Fuller (2001) found that SBNR respondents viewed spirituality not as a set of creeds, but instead as a sensibility, a particular mode of perceiving and responding to the world, which [cultivates] a mystical feel for God s presence in the natural world. Not surprisingly, people who identify as spiritual and religious feel the concepts are closely related and have positive associations with both, while spiritual not religious people feel the concepts are only distantly related and believe spirituality is superior to religion (Zinnbauer, 1997). SBNR respondents were also more likely to define religion in pejorative ways, such as a way to feel superior to others, a way to avoid responsibility for one s actions, or mindlessly accepting authoritarian tradition (Ammerman, 2013). If this synthesis between the two views is correct, then whether researchers find that Americans believe spiritual and religious are distinct concepts, or that the concepts are similar and overlapping, depends largely upon research design. Americans who are more conservative and/or belong to religious organizations may be expected to define religion and spirituality as overlapping, nested ways, or even use them synonymously. Americans who are more liberal and/or do not belong to religious organizations may be expected to define religion pejoratively and define spirituality positively. The majority of spiritual but not religious Americans fall into the latter group (Ammerman, 2013). Although the data used in this study does not allow examination of the meanings associated with the terms spiritual 9

15 and religious, other research supports the assumption that self-identified SBNR described here would see religion and spirituality as distinct concepts. Given the fuzziness of popular definitions of spirituality and religiosity, who, then, are the spiritual but not religious? Ammerman (2013) argues that the term spiritual but not religious does not represent a new religious orientation, but instead is a popular rhetorical device used to signal a deeply felt, unhypocritical faith. She believes those who adopt the term do so in order to create ideological distance from a religious other group, filled with people who lack a sense of a personal relationship with a deity, and especially those perceived as performing religious ritual out of tradition and/or obligation. The Spiritual but Not Religious Group Characteristics Whether spirituality and religion are separate or overlapping concepts, a further question remains; are spiritual but not religious people a group with distinct characteristics? Shahabi et al. (2002) present perhaps by far the most complete, yet somewhat dated, picture of SBNR Americans. Using the 1998 General Social Survey data, the researchers examined demographic, behavioral, and attitudinal correlates with a SBNR orientation. They included gender, race, age, region, city size, marital status, education, and family income as predictors. They found that those who are younger, white, currently married, and from the South are more likely to be SBNR than their respective counterparts. Shahabi et al. s study is most similar to this study; however, their data are 15 years old and call for an update given the extraordinary growth of the SBNR population in the past decades. Additionally, their predictors of SBNR 10

16 orientation are limited. Such predictors as, social class, work status, having children, sexual orientation, and non-religious socialization were not included. Overall, the literature on the spiritual but not religious as a group is very thin. Researchers disagree about how to define the SBNR, often without having a clear understanding of how the group is distinct from other religious categories. A more systematic analysis of the SBNR based on the latest nationally representative samples and more sophisticated methodology are called for. For the purposes of this study, I define the SBNR operationally, as those who call themselves spiritual but do not call themselves religious. I investigate the question of whether or not SBNR are distinguishable from religious people who are affiliated with an organized religion and from religious Nones including atheists, agnostics, and those who may believe in God but may not be affiliated with any religion. 11

17 HYPOTHESES To answer my second research question who is more or less likely to be SBNR holding other variables constant, I propose a number of hypotheses for testing. These hypotheses are organized in terms of the following categories of predictors: basic demographic variables, socioeconomic variables, locational variables, and family life variables. Demographic Variables Hypothesis 1: Males are more likely to be SBNR than females, all else being equal. Available evidence suggests that men are more likely to be atheists or religious Nones than women (Bryant, 2007; Chatters et al., 2008; Hayes, 2000; Kosmin, Keysar, Cragun, & Navarro- Rivera, 2009). SBNR is a subset of religious Nones, and so the demographic tendencies known to characterize Nones may be hypothesized to also characterize SBNR people. Hypothesis 2: Younger people are more likely to be SBNR than older people, holding other variables constant. Younger people are more likely to be disconnected from religion and spirituality than older people because of generational replacement associated with a changing and open social environment (Chatters, Taylor, Bullard, & Jackson, 2008; Hayes, 2000; Hout & Fischer, 2002; Hunsberger, 1985; Pew Forum, 2012; Roof, Greer, Johnson, & Leibson, 1993). In fact, many researchers believe that generational replacement is one of the major forces spurring religious disaffiliation (Hout & Fischer, 2002; Pew Forum, 2012; Voas & Crockett, 2005). Hypothesis 3: Blacks and other races are less likely to be SBNR than whites, respectively, controlling for other variables. One of the most well-documented social facts in the study of 12

18 American religion is that racial minorities are more likely to be religious than whites (Chatters, Taylor, & Lincoln, 1999). A recent Pew Forum (2012) survey finds that among the religiously unaffiliated, non-hispanic whites were overrepresented and that blacks, Hispanics, and Asians were underrepresented. Kosmin et al. (2009) largely agree, although they note that the percentage of religious Nones among Hispanics is rising. Religious self-identification is so strongly associated with minority groups, some scholars argue that the spiritual but not religious label marginalizes racial minorities (Wong & Vinsky, 2009). They argue that SBNR selfidentification, used as a marker of elite status, unintentionally discounts the social forces which drive racial minorities to identify with religious organizations. Whether this line of argument is accepted or not, there is no doubt that SBNR self-identification is associated with white Americans. Socioeconomic Variables Hypothesis 4: People with a higher level of educational attainment are more likely to be SBNR than people with a lower level of educational attainment, controlling for other variables. Education may be presumed to predict who becomes SBNR because it has been shown to have an effect on religiousness and spirituality in general (Kosmin et al., 2009). However, scholars disagree about the size and direction of the effect. Hayes (2000) finds that more educated people are more likely to claim no religion than less educated ones in at least five Western nations, including the United States. Pew Forum (2012) reports that religious Nones as a group tend to have only slightly higher educational attainment than as the general population and atheists and agnostics in particular tend to have more education than the general population. 13

19 Another group of researchers (Kosmin et al. 2009) finds no educational attainment differences between Nones and the larger population. Massengill and MacGregor (2012) argue that since the 1960s, the educational attainment of Nones has dropped relative to the general population and at present, lower educational attainment is associated with no religion. Although this study s examination of association between education and SBNR will not directly address the confusing state of research on Nones, it will contribute indirectly by examining a subset of Nones. Hypothesis 5: People who self-identify as lower class are less likely to be SBNR than those who self-identify as upper class, all other things being equal. Because both educational attainment and income are correlated with class, perhaps higher class can be expected to predict a SBNR orientation (Nakao & Treas, 1989). Further, religious organizational involvement is highly correlated with poverty and economic insecurity (Norris & Inglehart, 2004) and with lower educational attainment (Kosmin et. al., 2009). Therefore, I expect that people who selfidentify as lower class will be less likely to reject religion and call themselves SBNR. Hypothesis 6: Full-time workers are more likely to be SBNR than people who do not have full-time work, all else being equal. Although no research currently examines the relationship between work force participation and religion among SBNR Americans, some research does so for religious Nones (Hertel, 1988). Non-religious men are less likely to be fulltime workers than men who are religious; conversely, non-religious women are more likely to work full-time than religious women. However, controlling for gender, I expect that full-time workers will be more likely to be SBNR than non-full-time workers because lack of secure employment is associated with religiousness (Immerzeel & Tubergen, 2013). 14

20 Locational Variables Hypothesis 7: People in the Northeast, the Midwest, and the West are more likely to be SBNR than Southerners, respectively, holding other variables constant. Northerners are known to be more likely to be SBNR than other regions of the country, while Southerners have higher rates of religious organizational participation (Chatters et. al., 2008; Gunnoe & Moore, 2002). Pew Forum (2012) found that among the religiously unaffiliated, people from the West and Northeast were overrepresented and people from the South were underrepresented. Hypothesis 8: Urban dwellers are more likely than either suburban or rural dwellers to be SBNR, controlling for all other variables. Shahabi et al. (2002) found that Americans who lived in urban areas were more likely to be neither spiritual nor religious, but found no statistically significant effects associated with SBNR Americans. However, the correlation between urbanization and low religiosity is well-established (Barros & McCleary, 2003b; Ruiter & Van Tubeergen, 2009). Further, some researchers argue that spirituality is replacing religion more in urban environments than in rural environments (Kisala, 2004). Family Life Variables Hypothesis 9: People who are currently married are less likely to be SBNR than people who are not currently married, all else being equal. Married people are less likely to eschew religious and spiritual life than non-married people (Bock & Radelet, 1988; Chatters et al., 2008; Hertel, 1988; Pew Forum, 2012). In fact, Wuthnow (2010) argues that modern young adults tendency to delay marriage is one of the driving forces behind the general trend toward 15

21 religious disaffiliation. Similarly, Zhai, Ellison, & Stokes (2008) also contend that family disruption spurs religious disaffiliation, showing that children of divorce are more likely to be SBNR. Further, people who have ever divorced are more likely to become religious None than those who have never divorced (Lawton & Bures, 2001). Sociologists of religion debate whether this finding is indicative of age or life stages, but the correlation between current marriage and belonging to a religion is well-established (Sherkat & Ellison, 1999). Hypothesis 10: People who have no children are more likely to be SBNR than people with children, holding all other variables constant. Several studies indicate that religious Nones are overrepresented among the childless (Baker & Smith, 2009b; Hout & Fischer, 2002). However, no previous research tests the effect of childlessness on SBNR. Hypothesis 11: People who have had homosexual sex in the past year are more likely to be SBNR than those who have not had homosexual sex, all other things being equal. A survey of homosexual participants in a gay pride parade revealed that only 24.5 percent belonged to a religious organization, and respondents identified themselves as spiritual much more often than religious (Halkitis et al., 2009). Although this survey was non-randomized, nonrepresentative of homosexual spirituality and religiosity, other studies show that gay Americans often feel themselves alienated by organized religion (Buchanan, Dzelme, Harris, & Hecker, 2001; Morrow, 2003). Further, religious people are more intolerant on average of homosexuals than the general population (Shahabi et al., 2002). Hypothesis 12: People who were religious Nones at age 16 are more likely to be SBNR than people were not religious Nones at age 16, holding all other variables constant. Research confirms the common-sense notion that religious belief and affiliation, and especially lack of 16

22 religion, is largely inherited from parental figures via socialization throughout childhood (Myers, 1996; Voas & Crockett, 2005). Although the GSS does not ask respondents about their spirituality at age 16, it does ask about the broader category of religious orientation at age 16. Childhood socialization to reject religion may be a strong predictor of an adult s likelihood of being SBNR. 17

23 DATA AND METHOD Data The data for this study come from the General Social Surveys (GSS). The GSS is a random, nationally-representative survey of non-institutionalized U.S. adult population, which started in 1972 and continues through the present. The GSS dataset is ideal for studying SBNR Americans because of its representative nature and because of the broad range of topics included. In response to a surge in interest from researchers in spirituality and religiosity, the GSS included more questions on these topics in 1998, 2006, 2008, 2010, and This study only includes the surveys because they represent the latest data on these topics. SBNR demographics may be different than 1998 because of significant changes in the SBNR population in that time period (Kosmin et al, 2009). I pooled the 2006, 2008, 2010, and 2012 GSS data together in order to increase the sample size of the SBNR (N=4806) 1. I weighted the data in order to give each adult selected from each household an equal chance of being selected, so that the results can be generalized to the U.S. adult population. Dependent Variable Respondents were asked, To what extent do you consider yourself a religious person? Are you with the answer categories of very religious, moderately religious, slightly religious, or not religious at all. Respondents were also asked, To what extent do you consider yourself a spiritual person? Are you with the answer categories of very spiritual, 1 The total N based on the dependent variable is N=9033. The substantive N was limited only to respondents who had answered all questions used in my analyses. See the second footnote (p. 19) for more information. 18

24 moderately spiritual, slightly spiritual, or not spiritual at all. The dependent, spiritual but not religious, combines these two questions. Respondents who self-identified as not religious at all and very spiritual, moderately spiritual, or slightly spiritual were coded 1 spiritual but not religious. Respondents who claimed to be not spiritual at all and very religious, moderately religious, or slightly religious were coded 0 non-spiritual but not religious or religious but not spiritual. The total sample included 4,493 respondents, 579 of whom were spiritual but not religious. The process of operationalizing a spiritual but not religious variable raises more questions than answers. Researchers have struggled to measure religion and spirituality, developed conflicting methods of operationalization, and subsequently obtained radically different results, explored by Marler and Hadaway (2002). For example, the Gallop Organization asked respondents to choose if they were religious, spiritual, but not religious, or neither. Given this choice, 30 percent of respondents identified themselves as SBNR. If respondents are asked to choose between four spiritual and religious types ( spiritual and religious, spiritual but not religious, religious but not spiritual, or neither spiritual nor religious ), 19 percent of respondents will identify as SBNR. Other research has asked respondents to answer two questions with yes or no answers: Are you spiritual, and also Are you religious? Crosstabulation of results reveals that 18.5 percent are SBNR. Most Americans consider themselves spiritual and religious, and the terms are defined by this majority in ways which are not mutually exclusive, but overlap (Zinnbauer et. al. 1997). Therefore, Marler and Hadaway conclude that survey questions with inclusive categories, like the General Social Survey Likert scale questions, yield more reliable results than 19

25 survey questions with mutually exclusive categories which force choice between categories which most respondents believe overlap. If data is collected by asking respondents a question with a Likert scale answer, as the General Social Survey does, researchers are faced with yet another dilemma of how to define the spiritual but not religious category. If respondents are considered SBNR if they are very, moderately, or slightly spiritual and not religious, only 12.0 percent of respondents fall into the SBNR category. However, respondents can also be considered SBNR if they are very or moderately spiritual, and slightly or not religious. If data is analyzed using this method, 16.4 percent of GSS respondents fall into the SBNR category. Figure 1 shows how each of these methods of operationalization affects categories of religiosity and spirituality. 20

26 Figure 1. Variations in Spiritual and Religious Orientation Categories Dependent on Operationalization, General Social Survey, U.S. Adults Gabhart Operationalization Very Religious Moderately Religious Slightly Religious Not Religious Very Spiritual Moderately Spiritual Slightly Spiritual Not Spiritual 77.2% 3.5% 12.0% 7.3% Alternative Operationalization 2 Very Religious Moderately Religious Slightly Religious Not Religious Very Spiritual Moderately Spiritual Slightly Spiritual 48.8% 7.0% 16.4% 27.8% Not Spiritual Shahabi et. al. (2002) used the second method of operationalizing the SBNR category using the GSS, and duplicating their method was carefully considered. However, the first operationalization was deemed more methodologically conservative because it restricts the spiritual but not religious category by excluding respondents who are slightly religious. The 2 In this paper, the alternative operationalization was used only to conduct sensitivity analysis. 21

27 first method of operationalization better represents the qualitative differences between people of different religious and spiritual types, while the second method only represents quantitative differences in degrees of spirituality and religiousness. While I understand why previous literature included slightly religious respondents in their operationalization, I believe that only including not religious respondents is a stronger way to operationalize the spiritual but not religious category. Independent Variables Predictors are divided into four categories: basic demographic variables, socioeconomic variables, locational variables, and family life variables. Basic demographic variables include gender, age, and race. Gender was dummy coded so that males were 1 and females were 0. Age was reported in years as an interval-ratio variable. I created two dummy variables for blacks and other race and used whites as the reference category. Socioeconomic variables encompass education, self-identified class, and employment status. Education is an interval-ratio variable measured in years. For social class, I created three dummy variables for lower class, working class, and middle class, and used upper-class as the reference category. Employment status is a dummy variable with 1 for full-time work and 0 otherwise. Income was initially included in the analysis, but removed because its inclusion resulted in the loss of many cases. Locational variables include region of the interview and urban or rural residence. I created three dummy variables for Northeast, Midwest, and West, and used South as the 22

28 reference category. I also created two dummy variables for rural and suburban and used urban as the reference category. Family life variables include currently married, homosexual sex, and having no children. Currently married was dummy-coded with all currently married people coded 1 and 0 otherwise. Having no children was dummy coded as 1 and having any number of children was coded as 0. The question asked the gender of the people with whom the respondent had sex in the past year, and answers were combined with the respondent s gender in order to obtain a dummy variable with 1 indicating homosexual sex in the past year, and 0 indicating heterosexual sex in the past year. 3 Finally, the adult respondents indicated their religious preference at age 16. Those who were religious Nones at 16 were coded 1, all others coded 0. Limitations of Data A couple of limitations with the data should be acknowledged. First, spirituality and religiosity were self-reported, which is often the preferred method for understanding religion because of the difficulties inherent in letting religious organizations report data about their members (Kosmin & Keysar, 2006). However, this methodology creates the problem of social desirability bias (Hadaway, Marler, & Chaves, 1998; Presser & Stinson, 1998). I acknowledge the 3 The homosexual sex variable reduced the number of cases more than any other variable in the dataset because it is a secondary question after a filter question. The filter question was, How many sex partners have you had in the last 12 months? Only those respondents who had sex in the past year were asked, Have your sex partners in the last 12 months been. 1) Exclusively male, 2) Both male and female, or 3) Exclusively female? Despite the loss of cases, the measure of homosexual sex is considered extremely theoretically interesting and so was included in this study. 23

29 possibility that American social norms encourage respondents to overstate their religiosity or spirituality. Another major limitation with the use of this data is that religiosity and spirituality are presented as unidimensional concepts. Scholars of religion prefer a multi-dimensional index of religion which allows participants to specify belief, belonging, and behavior (Storm, 2009). Spirituality, as well, is best measured through an index which probes respondents perceptions of a higher power, emotions, commitments to spirituality, and motivations (Seidlitz et al., 2002). The General Social Survey does not probe either religiosity or spirituality in depth, but its inclusion of a broad range of demographic questions makes it ideal for this study. Methods and Analytical Strategies To answer the first research question, I first present descriptive statistics for the SBNR in comparison with the general population. To answer the second research question, I use logistic regression because the dependent variable is dichotomous. Four logistic regression models were created. Model 1 includes only demographic variables. Model 2 includes demographic variables and socioeconomic variables. Model 3 includes demographic, socioeconomic, and locational variables. Model 4 includes all other variables, plus family life variables. All models are shown in Table 2. I show the results of logistic regression based on the restricted operationalization discussed earlier because it is deemed more methodologically conservative. I then present the results of sensitivity analysis based on an alternative operationalization of the SBNR in Table 3. 24

30 RESULTS Who Are Spiritual but Not Religious Americans? Table 1 presents the descriptive statistics for the characteristics of the SBNR and the general population based on GSS data. Compared with the characteristics of the all respondents, as with religious Nones, the SBNR subsamples included slightly more males and were slightly younger than the overall GSS sample. SBNR respondents are 2.5 years younger on average than all respondents. Far fewer SBNR respondents were racial minorities (15.2 percent) than the overall GSS sample (23.3 percent). Only 76.7 percent of the overall sample is white, but 84.8 percent of the SBNR subsample is white. According to the descriptive statistics, SBNR Americans may be characterized as elite in terms of socioeconomic status. 4 SBNR respondents have almost a full year more of higher education than average GSS respondents. Further, more of them described themselves as upper or middle class, and fewer describe themselves as working or lower class. There was no substantial difference between the percentage of SBNR with full-time work and the percentage of the entire sample with full-time work. 4 Income was not used in this analysis because of the low number of valid responses. 25

31 Table 1. Descriptive Statistics of Spiritual but Not Religious (Standard Deviation in Parentheses), U.S. Adults, All Respondents 5 SBNR Male 50.9% (.500) 53.4% (.499) Age 40.5 (13.4) 38.0 (12.7) Race White 76.7% (.436) 84.8% (.373) Black 12.7% (.333) 6.8% (.252) Other 10.6% (.308) 8.4% (.280) Years of education 14.0 (2.80) 14.7 (2.67) Class Lower 4.3% (.204) 4.2% (.201) Working 49.3% (.500) 45.1% (.498) Middle 43.8% (.496) 46.6% (.499) Upper 2.5% (.175) 4.1% (.211) Full-time work 78.4% (.412) 78.2% (.412) Region South 36.8% (.485) 24.9% (.437) Northeast 16.0% (.366) 22.5% (.417) Midwest 22.7% (.419) 21.3% (.410) West 24.5% (.430) 31.3% (.464) Area Urban 57.2% (.490) 59.8% (.479) Suburban 32.6% (.469) 32.2% (.448) Rural 10.2% (.302) 8.0% (.271) Currently married 58.9% (.492) 46.2% (.499) No children 31.1% (.463) 49.1% (.500) Homosexual relations 3.5% (.184) 6.7% (.250) Religious none at % (.278) 21.5% (.411) N Only includes respondents with valid responses for all independent and dependent variables. 26

32 The GSS data also exhibit locational differences between SBNR respondents and the average population. The percentage of SBNR respondents who live in the South was 11.9 percent lower than the average. In contrast, the percentages who live in the West is 6.8 percent higher than the average, and the percentage who live in the Northeast is 6.5 higher than average. The Midwestern region shows the smallest difference between SBNR and average Americans, with only a 1.4 percentage difference. SBNR Americans appeared to be overrepresented among Westerners and Northeasterners, and underrepresented among Southerners. Differences between urban and rural residents were less dramatic, but noticeable, as 2.6 percent more SBNR respondents lived in urban areas than the national average, and 2.2 percent fewer SBNR respondents lived in rural areas than average. SBNR Americans seem to have different family demographics than the average population. Among the larger sample, 58.9 percent are currently married, but only 46.2 percent of SBNR are currently married. Similarly, only 31.1 percent of the total sample was childless, but 49.1 percent of the SBNR sample was childless, an astounding 18 percent difference. Only 3.5 percent of total respondents were calculated to have had homosexual sex within the past year, but 6.7 percent of SBNR respondents had homosexual sex in the past year. SBNR respondents were much more likely to have been religious Nones at age 16 than the average population. Only 8.5 percent of the total population claims to be a religious None at age 16, but 21.5% of the SBNR subsample was a religious None at age

33 Who Is More or Less Likely To Be Spiritual but Not Religious? Table 2 shows the results of four logistic regression models predicting SBNR orientation among respondents. The model fit statistics (-2 log likelihood and Model χ 2 ) indicate that Model 4 fits the data better Model 1, 2, or 3. Only one variable, age, changes in statistical significance from one model to another, and this change is examined in the Discussion section of this paper. Thus, my interpretations in this section focus on Model 4. Only one demographic predictor, race, is statistically significant at the.01 level or better. Neither sex nor age is a statistically significant predictor of being SBNR, controlling for all other variables. Blacks are 48 percent less likely than whites to be SBNR. People in the other racial categories are 40.3 percent less likely than whites to be SBNR. Among the socioeconomic variables, only education has a significant effect on the SBNR orientation, lending support to my hypothesis. For each year increase in education, the predicted odds of being SBNR will increase by 11.6 percent. Social class has no significant effect on the dependent variable, although all lower classes are somewhat less likely to be SBNR than the upper class. Full-time work does not make a statistically significant difference in SBNR orientation. The effects of locational variables are mixed. My hypothesis that people in other U.S. regions are more likely to be SBNR than the Southerners is partially supported. People from the Northeast, and the West are about twice, 1.9 times, and 1.6 times, respectively, as likely as Southerners to be SBNR. However, being from the Midwest is not a statistically significant predictor of being SBNR as compared with being from the South. Additionally, living in an urban, suburban, or rural area does not make a significant difference in SBNR orientation. 28

34 Table 2. Predicting Spiritual but Not Religious Orientation Logistic Regression Estimates (Standard Error in Parentheses) U.S. Adults, Model 1 Model 2 Model 3 Model 4 Predictor B Odds Ratio B Odds Ratio B Odds Ratio B Odds Ratio Male.125 (.090) (.092) (.092) (.095) Age (.003)*** (.004)*** (.004)*** (.004).998 Race (ref = White) Black (.172)*** (.174)*** (.177)*** (.181)***.520 Other (.158)** (.159)* (.163)** (.167)**.597 Years of education (.018)*** (.019)*** (.019)*** Class (ref = Upper) Lower (.329) (.333) (.340).705 Working (.249) (.253) (.256).679 Middle (.243) (.246) (.250).619 Full-time work (.112) (.113) (.116) Region (ref = South) Northeast (.134)*** (.137)*** Midwest (.131) (.134) West (.125)*** (.129)*** Area (ref = Urban) Rural (.171) (.175).972 Suburban (.100) (.103).909 Currently married (.112)***.659 No children (.116)*** Homosexual relation (.204)* Religious none at (.126)*** Constant (.150)*** (.410)*** (.421)*** (.437)*** -2 log likelihood Model X *** *** *** *** Degrees of freedom N *p < 0.05 **p < 0.01 ***p < (one-tailed test) 29

35 Family life predictors show an expected significant effect on the dependent variable. Consistent with my hypothesis, currently married people are 34.1 percent less likely to be SBNR. Also coinciding with my hypothesis, people with no children have about 63.4 percent higher likelihood to be SBNR as those with children. People who have engaged in homosexual sex in the past year are about 51.7 percent more likely to be SBNR. Non-religious socialization at the age of 16 has the greatest effect on the SBNR orientation as respondents who were not religious at the age of 16 are approximately 2.56 times more likely as those who were religious at 16 to be SBNR. Sensitivity Analysis Since the SBNR category type could be operationalized two different ways, I conducted a sensitivity analysis in order to test the robustness of my logistic regression results using the alternative operationalization of SBNR, the same predictors, and duplicate logistic regression models. The results are presented in Table 3. Comparing Table 3 with Table 2, it is evident that the results are very similar with a few exceptions. Just as in Table 2, age is significant in the first three models, but not the fourth. Sex, education, class, work status, region, metro area, marital status, parental status, homosexual sex, and religious None at 16 all retain their significance (or lack of significance) and their direction. All statistically significant odds ratios in Table 3 are equal or less than Table 2, indicating that the alternative operationalization yields similar, but less robust, results. Only one variable, race, changed significance due to changing operationalization. Black Americans are statistically significantly less likely to be SBNR in Models 1 and 2, but the 30

36 significance is lost in Model 3, when locational variables are controlled for. Conversely, Americans in the other racial category (not white and not black) are not statistically significantly different than white Americans in Models 1 and 2, and then they are shown statistically significantly less likely to be SBNR when locational variables are added. More analysis is needed to determine whether these effects are the result of controlling for region of the country, metro area, or both. However, it is clear that the alternative operationalization, with its inclusion of slightly religious and simultaneous exclusion of slightly spiritual Americans, dampened the more robust effects of race apparent with more restricted operationalization. If racial minorities are willing to claim that they are slightly religious, but unwilling to self-identify as not religious, then many more of them would be represented in the alternative operationalization SBNR type. Perhaps the lack of clear, robust racial effects reflect racial minorities reluctance to self-identify as rejecting religion. While the effects of these variables require additional testing, the main conclusions reached in Table 2 remain largely valid. Furthermore, I believe that the restricted operationalization of SBNR best measures the SBNR social category because it reflects qualitative differences in being spiritual while the alternative operationalization of SBNR only reflects quantitative differences in degree of spirituality. Hence, the results of the restricted operationalization are preferable those of the alternative operationalization. 31